View Full Version : mausoleum
01-11-2010, 04:34 AM
This is a beginning work in progress. I just started this, but I'm still interested in any suggestions etc.
01-13-2010, 01:38 AM
Here are a couple of new stills. Let me know what you think!
Another thing: Is it necessary to strictly follow the design principles of Romanesque architecture?
01-13-2010, 09:24 AM
I like what you've got so far. Are you using reference? If so, can we see some?
As for sticking to the architectural conventions, that's entirely up to you. If you want to change something, do it. As long as the end result looks good and the style is recognizable, it doesn't matter. If it were me, I'd stay true to the design principles as closely as possible, but then again I'm an architecture junkie and wouldn't be happy changing it, but that's just me.
Are you planning on getting this into an engine? Going for just the asset, or a complete scene? Either way, keep at it!
01-15-2010, 02:27 AM
I've been focussing on getting the column right. At first my polycount was way too high (from 76438 to 5672), so I rebuilt it. Here are the wires:
Suggestions? I will be reworking the capital next. Variation in the columns will probably be done through the textures btw.
Oh... And the references I have:
Here are the two that I'm using to base the overall structure of the mausoleum on:
Here is what I'm mainly going to base the door on:
The column capitals in this image are what I copied to make my capitals. Although this is the kind of capital I really want:
01-15-2010, 09:49 AM
Okay, so you say that your polycount was too high, so you rebuilt it. I assume, then, that you're building the low poly first? I'm also guessing that you're going for a current-gen specification model, using normal maps?
The reference images are good - plenty to work with, and some good textural detail as well as structural. Be mindful of scale when using different reference images, though - the detail on the door you've chosen looks good, but that door appears considerably larger than the one in the top image, which looks to be your primary reference image. Recreating the detail exactly may not translate well to a smaller door, or fit well with the aged, weathered look of the top image.
01-15-2010, 11:49 PM
Actually this is the high poly model I'm working on right now. It's just that the column(s) had way too many polys. I'll post more stuff soon. Thanks for your comments!
01-16-2010, 02:48 PM
Here's the latest version of the column capital with some leaves. Let me know what you think!
01-17-2010, 01:39 PM
I made some progress with the door. Here are a couple renders and the wireframe:
01-19-2010, 03:20 PM
Here is my latest progress. If I could get feedback on the rendering quality as well as whether or not I'm near the end of creating the high poly model that would be great.
01-19-2010, 04:35 PM
01-19-2010, 04:39 PM
The modeling by itself isn't bad, although it is still very basic. What exactly is this model for? Right now I would assume some kind of architectural render since it's not really built for normal mapping.
Either way, the presentation is still very poor. I'm not as familiar with Maya, but I am with Mental Ray. You should be able to use a simple daylight system to light this model and have it look 100 times better than it does now. The scene below was built in 2 minutes and renders in about 30 seconds.
Setting your model up and presenting it in a professional way will get you more critiques too.
Lastly, there isn't really a need to post a render or reference photo greater than 800x600 here. It's a bit annoying to have to scroll all around the page to see these giant images.
01-20-2010, 03:16 PM
I have some new renders. Plus I added bricks to the walls...
01-22-2010, 01:56 PM
I just spent quite a bit of time rendering some more. I've decided that I'm going with ambient occlusion renders. I know there are multiple ways of presenting work, AO renders are the simplest and most effective for me presently. I also added some detail to the two rectangular columns near the entry way. I'm still incorporating some other suggestions as well.
I think these may be a little overblown. I'm experimenting with the number of final gather rays being used.
01-24-2010, 03:03 PM
Here are the latest images:
Unless there are any glaring issues, I think I'm going to stop adding to this high res model, do some optimization (finding a turbo-smooth type function in maya), and then move on to the low res model. I'll add wear and imperfections in the walls and columns with the normal map. I'm not aiming for things to be absolutely perfect, just satisfactory enough. I'm not sure if I mentioned this, but this is the first piece ever for the first version of my portfolio.
Let me know what you think.
01-30-2010, 01:20 PM
I have finished the high poly model. I'm now trying to add further details in Zbrush, which I just got a couple days ago. I've imported the model as an OBJ file and have noticed that the mesh is somewhat distorted. Does anyone know what this could be?
Here are some screenshots:
01-30-2010, 06:00 PM
you need to eliminate Triangles , Zbrush doesn't like them
01-31-2010, 04:55 PM
You need to slow way down on this project, and get some basic training. Zbrush is not just some 'detail generator.' Learn how to use the software and practice for at least a few weeks before you attempt to work on this. Also, it's highly unlikely that you will simply export your entire mesh into Zbrush. Doing so shows a complete lack of knowledge for a 3d art pipeline of any kind. I may have mentioned this before, or maybe not, but you can generate as many portfolio pieces as you want, but if you don't do them right they are useless.
Major training is needed here. I strongly suggest you find some good DVD's, for modeling, lighting, and whatever else you are interested in and spend a few months with them.
01-31-2010, 05:31 PM
Also, it's highly unlikely that you will simply export your entire mesh into Zbrush. Doing so shows a complete lack of knowledge for a 3d art pipeline of any kind. I may have mentioned this before, or maybe not, but you can generate as many portfolio pieces as you want, but if you don't do them right they are useless.
Since it's a single asset, I don't see anything wrong with exporting the entire mesh into ZBrush, as long as he imports individual elements as separate sub-tools to make it easier to work with. Even if he didn't, it's not like it's impossible to achieve good results this way, just a little more challenging and probably not the most efficient way to work.
Most people learn most effectively by doing, so what's wrong with getting stuck in to a project like this? ZBrush can be awesome for environment work, and surely it's better to experience an application by working on a real project with specific goals in mind.
Your belabored point of 'doing things right' is pretty redundant. If it looks good and meets the polygon / texture budget, job done. Nothing in CG is ever 'useless' - mistakes are far more effective learning tools than preachy, pretentious posts with sweeping generalizations and little in the way of constructive criticism. Saying 'you're doing it wrong' means nothing unless you back it up with either suggestions on how to improve the process, or at the very least an explanation as to why they're doing it wrong.
02-01-2010, 01:44 PM
Thanks for the input, I have been speaking to the OP through PM but I should have included more info in the post. I put in the effort to say something, but not necessarily something constructive.
I made the assumption that he was just getting started with Zbrush, so what I meant to suggest was to try a few test objects first to get the hang of it. Most people recognize that Zbrush has a unique workflow associated with it, so perhaps scuplting a single small column or rock as practice would be beneficial before tackling the main project. Simply importing the whole building at once is going to be a very inefficient way to create detail. He even went so far as to import all the metal bars and molding, while my instincts tell me these things are not setup properly for sculpting.
The reason I stress learning to do things right is mostly to help avoid the trap of being too "self taught." A lot of people get their start following random tutorials on the web, and while helpful, when learning this way it is easy to miss the big picture. I don't believe he has a polygon count or texture budget, or a realistic way to generate these without more knowledge of a game asset production pipeline. Doing simple pre-production and some planning before you dive into a model is always a smart idea.
One good thing he is doing is to post work here for feedback, however CGtalk seems to have a relatively small section devoted to game art. Try posting this asset on gameartisans or polycount if you want to get some additional feedback.
Lastly, make sure to lose the attitude of "I don't care if my models are excellent, they just need to be good enough." Whether or not they will end up being good is up for debate, but don't sell yourself short up front. It also gives others the impression that you don't want or won't listen to feedback, so they might not bother to give any.
I've seen a lot of artists simply put out piece after piece without any signs of improvement or awareness, so my point in posting here was hopefully to nudge him in the right direction. He has a great attitude and what seems to be a good interest in this stuff, so hopefully my efforts weren't totally in vain. I believe the way to get good at something is to spend time practicing it, but you also need to spend equal time watching professionals and learning from them. Working on DVD's and professional tutorials is a great way to do this.
02-01-2010, 06:24 PM
I just saw that you did in fact post this over at game-artisans. So good job for that.
02-02-2010, 11:20 PM
Apparently I had to hit "double" in the display preferences in ZBrush in order for my model to show up properly, even when I made sure that all my normals faced outward in Maya. I tested the model by bringing it back into Maya and everything was fine.
Even though I like my overall model, its geometry needs improvement, as has been pointed out.
Nonetheless I wanted to experiment a little in adding wear to the column, just to see what it was like. Keep in mind these were just quick attempts, I know they are lacking and are a little strange.
Since I'm getting more familiar with topology's importance, I added to the mesh of the column. I know this needs to be subdivided, along with the rest of the model.
Finally, there has been concern over the detail that's actually part of the model. I prefer to model things as much as I reasonably can since I'm not exactly comfortable (at this point) in creating the detail in Photoshop.
Comments appreciated as always. Thanks!
02-03-2010, 05:49 PM
I've now started the somewhat laborious process of adding supporting edges to the model so it can be smoothed and allow for a better mesh. I can see now that this is something I should've done from the start... Since using Zbrush is definitely part of my personal process, I additionally try to make sure that there are no holes or n-gons in the model as well. This initially might be a little time-consuming, but I really like the amount of detail I can get with the brushes etc.
I'm going through element by element with the model and fixing the geometry. I've done the geometric pattern above the door.
An experiment with adding rust and damage:
Comments as always are greatly appreciated. Thanks.
02-23-2010, 10:33 PM
I was away for awhile... I have finished the model in Maya. It has supporting edges etc so it can be smoothed properly. I was experimenting with bringing in parts of the model as seperate tools into Zbrush, then as one whole mesh split using the "group split" command (caused it to crash)... I'll be learning about Maya/Zbrush workflow in a little while. I just wanted to check in, so to speak.
For good measure (perhaps redundantly) here's a shot of the model:
02-28-2010, 05:44 PM
I've started working on the brick walls. Comments?
02-28-2010, 05:44 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.